Contextual service switch

As a user experience designer at Rewe Digital, my focus is primarily on analytical and continuous development. This means that I always aim to combine user needs with the business needs of the company. Short-term successes are worthwhile, but long-term successes should be the goal.

This case study is an exciting example of how we improve business KPIs by enhancing the user experience.

Rewe Digital is a subsidiary of Rewe, one of the largest retail companies in Europe (turnover in 2020: around 75 billion.) Rewe Digital’s task is to develop digital services and products within e-commerce, fulfillment, and the digitalization of stationary markets. As User Experience Team we work within an agile environment in different Tribes (eCom, Content/stationary, and fulfillment).

Mockup – – Service switch vom Delivery to Pick-up
Mockup – – Service switch vom Delivery to Pick-up

Rewe offers different services

REWE offers various digital services which, depending on the region, are either available at the same time or only separately. These services are the delivery service (a delivery driver delivers the order at an agreed time) and the pick-up service (pick-up at a Rewe store). These services serve very similar customer needs, which, however, differ in their characteristics. Both services are used for weekly shopping but have different conditions. The REWE delivery service has a minimum order value of € 50, and there is no minimum order value for the pick-up service.

Problem Statement

As a User, I want to know when it is possible to order my weekly grocery shopping in service with a less high minimum order value and switch to that service without losing my basket.

AB-Testing: Switch services on the basket page

Our goal is to offer our users an easy way of shopping at any given moment in their journey. In the delivery service, we have to offer our customers a relatively high minimum order value. For some users shopping online is interesting, and they start to put together a shopping cart, but we noticed that they cancel their purchase at the basket. Since we have a significant overlap in the service areas and know that we also offer many similar products in the pick-up service as in the deliver serivce, we want to discover whether users would complete their purchase in another service.

Delivery ServicePick-up Service
Minimum Order Value 50 €No Minimum Order Value
Approx. 16.000 Products Approx. 8.000 Products
A delivery driver delivers the order at an agreed time. Customers pick up their already compiled shopping cart at a Rewe store.

AB Test 1: Lean Testing with Optimizely

User Flow – Switch service from Delivery to Pick-up

With the first AB test, the aim was to validate the thesis in a lean test. To do this, we used Optimizely to implement the code and track the results. We knew how we implemented the switch of the services was not perfect since users had to leave the basket page to change the service.

Our thesis for this test: Those interested in Deliver Service who do not reach the minimum order value of 50 € are interested in the pick-up service (click at least X% on a link to the AS in the shopping cart).
Result: In the test, we found a significant intersection of users for whom such a notice is relevant and who use this service change at this point. With this initiative, we were able to increase the number of orders in the pick-up service without reducing the conversion rate in the delivery service.

“The minimum order value of 50 € was not reached.”“The minimum order value of 50 €
was not reached.”

“Complete shopping with the
pick-up service without a minimum order value and pick up the goods in the REWE store.”
Test Variant 1) Control and 2) Treatment. Control: No Service Switch possible. Treatment: Service can be switched

Implementation & Next Steps

After the positive result of the AB test, we have further developed the notification in different versions. We could streamline the notification in terms of content and visual appearance. However, the original flow is still not perfect, as users navigate via a separate domain and therefore leave the shopping cart and hence the checkout.

Before users enter the shopping cart, there is the situation that users are asked to select the service from which they would like to orderImplementing this selection in the shopping cart would greatly simplify this flow and therefore offers a good chance that the conversion rate will improve again. Consequently, we had to wait to implement the so-called “Market Chooser” before starting the next test.

Iteration 2Iteration 3 (with Market Chooser)
“The minimum order value of € 50 has not been reached.
Switch to the REWE pick-up service and presume without a minimum order value:
Switch to the pick-up service
“The minimum order value of € 50 has not been reached.
Shopping under 50 € minimum order value:
Switch to the pick-up service
Change from Delivery to Pick-up within the Basket using the “Market Chooser”

Test 2: Button instead of a Text Link

After the Market Chooser was implemented, we were ready to proceed with testing. We already know that changing the service in the shopping cart is a relevant function for the user because of the minimum order value. We now have to find out whether this effect could be used even further by a visually enhanced design. We tested two different styles, and we also found a positive impact here and decided to keep the style of our treatment variant.


“Switch to the pick-up service and make purchases with a minimum order value of less than € 50.
Text link: Switch to the pick-up service

“Switch to the pick-up service and make purchases with a minimum order value of less than € 50.
Button: Switch to the pick-up service


Testing different variants against each other are one of the essential characteristics of the continuous development of a website, and it is a process that we use very often. I think this example is exciting because the tested element took a very non-linear development path.
From the user’s perspective, a well-known assumption is valid. If we offer users relevant information and features at appropriate points of their journey, they will acknowledge it, and essential KPIs will find an uplift. This test was also the starting point to further force this effect of switching between the services.

%d bloggers like this: